Autism is a severe neurodevelopmental disorder defined by social and communication deficits and ritualistic-repetitive behaviors. Encouraged by reports of a high heritability and a sibling risk ratio > 50, a substantial number of linkage and positional cloning studies have now been conducted. Employing primarily the affected sib pair paradigm, these studies have met with limited success, generally identifying modest linkage signals covering broad genomic regions that include very large numbers of candidate disease genes. The common responses to this dilemma (to increase sample sizes and marker densities) have met with numerous practical and theoretical obstacles. In this application, we instead propose a strategy using multigenerational, extended pedigrees with multiple individuals affected with autism, as well as milder but qualitatively similar, genetically-related phenotypes (endophenotypes) in non-autistic relatives. Linkage studies in extended pedigrees are recognized to have substantially more power than those employing affected sib pairs and have been successful in other complex disorders (1). Although large pedigrees in autism are rare, through our work over the past 10 years, and through collaboration with colleagues, we have identified 20 potential extended, multigenerational pedigrees in the U.S. and Canada. In this application, we propose a detailed phenotypic study (including measurement of quantitative and qualitative neuropsychological and behavioral characteristics in autistic individuals and their non-autistic relatives) of twelve of these pedigrees as part of a whole genome linkage study of autism and related phenotypes. These twelve pedigrees will include approximately 60 individuals with autistic disorder and 60 individuals with the broad autism phenotype. We have previously demonstrated that incorporating additional phenotypic information substantially increases power to detect linkage. In this application, we also propose novel analytic strategies, developed by our research team, expressly to allow efficient and flexible incorporation of the phenotypic measures proposed here, including joint linkage analysis of dichotomous and quantitative traits. [unreadable] [unreadable] [unreadable] [unreadable]